Many years ago, I read a short story by legendary Sci-Fi author Arthur C. Clarke called “The 9 Billion Names of God.” The story revolved around an order of Tibetan monks, who had determined that there were 9 Billion possible variations on how to write the name of God and, that once this task had been completed, mankind’s reason for existence would be fulfilled.
Not surprisingly, the monks were finding that this task was taking a long time to achieve by hand. Leveraging the newly available technology of the time, they commissioned an IBM mainframe ( this was 1953 ), and 3 months later the job was done.
It was after the admittedly sceptical IBM engineers had decommissioned the computer, and were heading back down the mountain trail to the nearest airport, that they suddenly notice that “overhead, without any fuss, the stars were going out.”
I was thinking of this story while I was visually inspecting a new client’s email list prior to broadcast, and starting to wonder precisely how many ways there are that people can mis-spell Hotmail ! So I ran the file through our suite of email address hygiene applications, and the answer would seem to be close to 100 – and those are just the ones where I had at least 90% confidence that Hotmail is the domain that they were actually intended to represent.
There was “htomail.com”. And “hotamil.com”. And “homtail.com”. And “hotmali.co.uk”. And “ohtmail.com”. And . . . – the list is ( nearly ) endless. And that’s before we even start to consider the likes of “btinterent.com”, “ayhoo.co.uk”, “ntlwolrd.com”, “tisclai.co.uk”, and “gogolemail.com” ( all real examples, by the way ! ).
Sometimes these errors are made on purpose, by people who are responding to a call to action ( “give us your email address and you can download our 2 for 1 voucher” ). Consumers are increasingly savvy, and want the reward without exposing themselves to a blizzard of e-marketing activity. This could point to issues with engagement – another topic in its own right. However, these errors are more often the result of genuine errors where people either can’t type or can’t spell ( or both – as an aside, this has been recognised by the developers of www.fatfingers.com which finds mis-spelled entries on eBay such as “nitendo wee” where bid activity is – not surprisingly – much lower, so the chance of a successful bid is correspondingly higher . . . ! ).
Anyway, back to “hootmail” – oops ! While it’s ( almost ) funny, this also represents a serious problem for e-marketers. Bounce activity is one of the core metrics that feeds into the calculation of a sender’s reputation data. While the vast majority of these addresses will be filtered out after the first time that they have been broadcast to, you’ve still got to generate the bounce notifications first ! e-Marketers have traditionally dealt with this by using a separate IP address for the welcome email / first broadcast, but the increasing use of domain-based reputation means that this isn’t a perfect solution either.
Because of this, it’s starting to place a premium on making sure that the address is being captured correctly in the first place. That means using a technique such as double-entry, so that if our fat-fingered friends get it wrong the first time, it will be flagged up when they enter the address correctly ( hopefully ) the second time. Alternatively, to use a validated opt-in process that generates a confirmation email and only activates the account once the confirmation email has delivered successfully.
However, even then you are not completely out of the woods, because some of the incorrect spellings are actually valid domains, a fact that major ISPs such as Hotmail have recognised, and which they are now monitoring as a new form of spam trap. What makes this particularly difficult is the fact that these records won’t usually generate a bounce notification, so you could end up perpetuating the problem by sending to them many times over – and ending up with a mail block every time. So the only effective way of dealing with these addresses once they are on your database is to monitor your response behaviour, and screen out all non-responders on a regular basis ( which is actually part of the best practice that you should be applying anyway ! ).
However, a better way of dealing with this problem is to pre-empt it by pre-screening your data prior to broadcast. As a business, we have developed an email address hygiene routine that applies fuzzy-matching logic to test supplied domain names against known valid domain names, and to then filter out all matches where there is statistical confidence about the likelihood of the supplied domain being an incorrectly spelt version of the valid domain. The match rates aren’t massive – on average, about half of one percent – but that’s still 50 records in a list of 100,000 addresses, which based on our experience is definitely more than enough to trigger a block against you if they are not removed.
So the key learning here is that the bar for good email data hygiene is being set increasingly higher. Standard bounce management processes are certainly no longer enough – validation of email addresses at point of capture, pre-screening to remove known errors, and recency analysis of open and click behaviour are all now playing a crucial role in the delivery of successful email broadcasts. Implement them now, and the stars will continue to shine on your e-marketing program !
Tags :













You must be logged in to post a comment.